Specifications

Chapter 18: Improving recognition accuracy
Notes
n Bydefault,theamountofacousticdatasavedislimitedto180minutes(the
maximum).YoucanreducethisamountontheDatataboftheOptionsdialogbox.
n IfyouattempttoruntheAcousticandLanguageModelOptimizerbeforeyouhave
trainedanycorrections(evenifyouhavecorrectederrorswithouttrainingthe
corrections),orbeforeyouhaveperformedanyadditionaltraining,youwillseean
errormessagesayingthatthereisnotenoughdatafortheAcousticandLanguage
ModelOptimizertoprocess.
n Trainingmultiplewordsorphrasesratherthantrainingindividualwordsisgenerally
moreeffectiveatincreasingaccuracy.Mostpeoplehaveatendencytoover-
enunciatewhensayingindividualwordsasopposedtoacompletephrase.
n TheAcousticandLanguageModelOptimizerautomaticallymakesabackupofthe
oldProfilebeforeitbeginsmodifyingthem.YourProfileisautomaticallyrestoredif
youclickCancelbeforetheAcousticandLanguageModelOptimizerisfinished.

242